The experimental study towards the mutual adaptive interface has been perfomed in order to validate the effectiveness of the adaptive mechanism based on the estimated human state. The Mental Work Load(MWL), which is estimated bnscd on the physiological meascli'es. has been taken a s a representative index of the subject and used to control the task dificulty in the example t;Lsk(X-window-bascd game, called X-jewel). The results of the laboratory experimenls showed the positive effect in the performance scwre. which implies the cffectivencss of the adaptive mechanism in the present task domain.
. INTRODUCTIONAlong with the incrcasing dcgrce of the automation in the modcm twhnological systems, the role of the human operator has shirtcd from the manual operation to the supcrvisory task. Although thc work load during the normal operation has reduccd significantly, the operator is required to pcrform critical task at thc case of system failure, when the automatized system is not capable of handling the situation. In the aviation society, thc problem of automation is discussed undcr the name or "Human Centered Aircraft Automation"[ 11. It has been pointed out in [l] that the following attributes are quite important in the course of progrcssing automation. Accountablc / Subdinate, Predictable / Adaptive, Comprehensive / Flexible, Dependable / Informative, Error-resistant / Error-tolerant. It is pointed out that some of these factors have some trade-off and some compromise is required depending on the situation.In the present study, we have set our focus mainly on the adaptability in the human-machine interaction among above mentioned attributes required for reliable automation. In [2], we have proposed the basic concept of the mutual adaptive interface, with two outstanding features; Human state estimation based on the physiological measures and Use of feedback controller. The proposed system aims at performing multi-level adaplation based on the cstimatcd human states taking thc dynamic naturc of thc human being into account. In othcr words, thc present study is an attempt to control the trade-off among the attributes required according to the faced situation and the state of the human operator.Although the validity of thc human state estimation has becn confirmed through the laboratory cxperimcnts[31, the effectiveness of the adaptation itself has not been validated in the prcvious work. In the prescnt paper, the experimcntal study towards the mutual adaptive interface is described with the emphasis on the validation of the adaptation mechanism. In chapter 2, thc basic concept of the mutual adaptive interface is revisited briefly and the purpose of the experiment is described. In chapter 3, the experimental setup and the content of the experiment are depicted followed by the results. After the discussion of the results, the papcr will be concluded by giving the future direction of the present work.